Unlocking prison performance. Elizabeth Crowhurst Eleonora Harwich

Unlocking prison performance Elizabeth Crowhurst Eleonora Harwich April 2016 #reformprisons

Unlocking prison performance Elizabeth Crowhurst Eleonora Harwich April 2016 Reform Reform is an independent, non-party think tank whose mission is to set out a better way to deliver public services and economic prosperity. Our aim is to produce research of outstanding quality on the core issues of the economy, health, education, welfare, and criminal justice, and on the right balance between government and the individual. We are determinedly independent and strictly non-party in our approach. Reform is a registered charity, the Reform Research Trust, charity no.1103739. This publication is the property of the Reform Research Trust. 1

Acknowledgements Advisory board Reform is particularly grateful to the expert advisory board who supported the authors on this project and provided feedback on early drafts of this paper. Ian Brunton-Smith is a Senior Lecturer in Criminology at the University of Surrey. He is the founder of the Surrey Crime Research Lab, member of the Royal Statistical Society Social Statistics Committee and sits on the editorial board of the British Journal of Criminology. Ian also worked for the Ministry of Justice on the Surveying Prisoner Crime Reduction longitudinal cohort study of prisoners. Manon Durand is a Research Economist at the Assistance Publique-Hôpitaux de Paris. She has worked for the French Ministry of Health and for the French Department of the Commissioner-General for Sustainable Development. She specialises in health economics and public policy evaluation. Manon has extensive experience of using a range of quantitative methods including Data Envelopment Analysis which is utilised in this paper. Kevin Lockyer worked for nearly 25 years in prisons and probation, as a prison Governor and as a senior civil servant in the National Offender Management Service. After a spell as a Director at the crime reduction charity Nacro, he now runs his own consultancy business and has supported a number of large private and voluntary sector organisations during the Government s Transforming Rehabilitation programme. Further thanks The authors would also like to thank Amy Finch, Oliver Lodge, Jerry Petherick, Charlotte Pickles and Marijn Verschelde for their helpful comments on drafts of this paper. The arguments and any errors that remain are the authors and the authors alone. In addition, the authors would like to extend a special thanks to Maximillian Hall, Nicky Rogge, Richard Simper and Marijn Verschelde for kindly providing their R code which served as a basis for this analysis. This code uses the Benchmarking package developed by Peter Bogetoft and Lars Otto. 1 2 1 Lars Otto and Peter Bogetoft, Benchmarking: Benchmark and Frontier Analysis Using DEA and SFA, version 0.26, 2015.

Contents Executive summary 4 Introduction 7 1 A review of the estate 9 1.1 The population 11 1.2 Spending 13 1.3 The estate 14 1.4 Cost vs quality: assessing performance in the long term 16 2 Measuring performance in prisons 18 2.1 Why outcomes count 19 2.2 Current methods 21 2.3 The way forward 24 3 A new model for success 26 3.1 Measuring best practice 27 3.2 Selecting meaningful metrics 28 3.3 Model constraints 30 4 Prison rankings 36 4.1 Efficiency 37 4.2 Effectiveness 54 4.3 Linking the chain 61 Conclusion 66 Technical appendix 67 Bibliography 82 3

Executive summary Increasing productivity remains a key challenge for government. This is particularly true of the public sector where official estimates indicate that productivity has grown by just 0.1 per cent per annum between 1991 and 2013. 2 As the Chancellor has argued, without sustained productivity improvements the United Kingdom will fail to maintain living standards and achieve economic growth in the longer-term. 3 Current productivity measures, however, are fundamentally flawed. 4 For the most part they remain sector-level analyses focussed on the volume of outputs produced, rather than the quality of the services or their value to citizens. Despite the pressing need to find productivity gains, little is known about how individual public institutions are performing, hindering the ability to learn from best practice and maximise value for money. This problem is very evident in the prison service. Official estimates use the number of prisoners held as the single output measure, which fails to take into account either the conditions for offenders or the quality of rehabilitative support provided. Government measures of prison performance also fail to take advantage of reoffending data. What matters is not being effectively measured. Hence, whilst the significant savings delivered over the last Parliament may indicate improved productivity in the prison estate, increases in violence, overcrowding and self-harm show deteriorating outputs but not ones that current productivity estimates capture. 5 In addition, reoffending rates have barely changed in a decade. 6 These factors have led the Justice Select Committee, amongst others, to question whether a focus on short-term cost-saving measures have, in some prisons, resulted in poorer longer-term performance. 7 It is therefore clear that a new performance measurement framework is needed: one which captures both how prisons spend their money to provide a safe environment and whether they improve the life-chances of offenders released from their care. Encouragingly, the Government recognises this: in a landmark speech on prison reform the Prime Minister announced the creation of new prison league tables which would balance the need to dip test performance at a given point in time with measuring longerterm outcomes. 8 To date, however, the method and framework for producing these remain unknown. This paper seeks to fill this gap and lays out a new model for performance measurement. Through ranking a group of comparable prisons against a frontier of best practice Reform also aims to identify high and low performing prisons and thus the scope for improvement across a number of metrics. Firstly, the analysis considers prison efficiency by evaluating whether prisons spend wisely, keep prisoners and staff safe and promote rehabilitative activities. Secondly, an evaluation of prison effectiveness looks at how successful prisons are at reducing reoffending and supporting prisoners into education, training, employment and accommodation on release. Taken together these measures allow an assessment of performance in the short and long term, and most importantly help determine whether individual prisons are delivering value for money to taxpayers. There is, however, a significant need for improved data availability and quality. The success of any performance model hinges on the integrity of the data used and as this 4 2 Office for National Statistics, Sources & Methods for Public Service Productivity Estimates: Total Public Services, 2016. 3 HM Treasury, Spending Review and Autumn Statement 2015, 2015. 4 Elizabeth Crowhurst, Amy Finch, and Eleonora Harwich, Towards a More Productive State (Reform, 2015). 5 Ministry of Justice, Safety in Custody Quarterly Update to September 2015, 2015. 6 Ministry of Justice, Proven Re-Offending Statistics Quarterly Bulletin April 2013 to March 2014, England and Wales, 2016. 7 House of Commons Justice Committee, Ninth Report of Session 2014 15, Prisons: Planning and Policies, HC 309 (London: The Stationery Office, 2015). 8 David Cameron, Prison Reform: Prime Minister s Speech, Speech, (8 February, 2016).

report argues, current data is both lacking and poor quality. In particular, publically available financial data is woefully inadequate. Whilst some financial data can be obtained on public prisons via a Freedom of Information request, the Ministry of Justice do not even collect this data for private prisons. This limits the ability for contractors to be held to account, but also for lessons to be learnt where private prison places are cheaper. For example, at privately run HMP Oakwood the cost per place is 12,210 per annum compared with an average of 21,382 for Category C prisons. 9 Within these constraints however, Reform s analysis of 40 Category B and C prisons shows that, across both the four efficiency and two effectiveness indicators, there is significant variation in performance. Closing the gap between the best and worst performing prisons therefore presents considerable opportunity to both realise savings and improve outcomes. The analysis also shows that the most efficient prisons are not necessarily the most effective, reinforcing the potential need to make trade-offs in the short and longer-term. Few prisons are able to transform good prisoner living conditions or high levels of resettlement provision (for example accredited courses and drug treatment) into improved life chances for offenders on release. Data availability and quality prevents further analysis of this, but the Ministry of Justice should prioritise further examination of those prisons which are able to buck the trend and perform well against both measures. The forthcoming Prison Reform Bill provides an opportune moment to address these issues and bring greater clarity and transparency to prison performance. This is an essential step for a Justice Secretary committed to reforming the prison estate to deliver better outcomes for prisoners. 10 Greater transparency and a focus on outcomes are also essential for increasing productivity and delivering value for money. This report aims to provide a helpful step towards this goal. 9 Ministry of Justice, Cost per Place and Cost per Prisoner by Individual Prison Establishment 2014-15 tables, 2015. 10 Michael Gove, The Treasure in the Heart of Man Making Prisons Work, Speech, (17 July, 2015). 5

Unlocking prison performance / Executive summary Summary of recommendations 1. The Ministry of Justice should collect and publish data documenting broken down expenditure patterns for both private and public prisons. This should include, at a minimum, the amount of funds spent on payroll, building maintenance, prisoner training (including education and industries) and drug testing. 2. The Ministry of Justice should where possible increase the diversity of reoffending data publically available (including severity and time to failure at an institutional level) to support the identification of strategies to reduce offending behaviour. 3. The Ministry of Justice should create a baseline of predicted reoffending at a prison level. This will enable a better understanding of performance by taking into account the impact of external factors on outcomes. 4. The Ministry of Justice should introduce a measure of prison performance which better encapsulates prisoner activity. This should include, at a minimum, time spent on education, industry, accredited programmes (taking into account course completion rates) and any hours spent as part of peer mentoring schemes. To ensure governors and prison staff are not incentivised to provide activity for activities sake through tasks which are unlikely to develop skills or promote rehabilitation, a framework should be established which lays out which activities can be included under the new measure. 5. The Ministry of Justice should instruct prisons to collect data on the number of prisoners in denial of their offence. This should be ascertained through a combination of prisoner input and staff assessment based upon conduct both during the prosecution process (as documented by Her Majesty s Courts and Tribunals Service) and whilst incarcerated. 6. The Ministry of Justice should collect data on the number of visits received by prisoners as a proxy for family ties. Once the digital prisons programme has progressed further, time spent on the phone or video-conferencing family should also be included in this metric. 7. In order to develop a more rounded performance framework the Ministry of Justice should include qualitative evidence of prison performance. To support this independent bodies such as the Her Majesty s Inspectorate of Prisons and Independent Monitoring Boards should also ensure more conformity and detail in their reports. Taken together these measures will enable greater transparency and allow more comparative research to be undertaken to understand of the drivers of prison performance. 8. The Ministry of Justice should set minimum targets for the provision of substance misuse courses (specifically within Category B and C prisons) a practice which is currently employed for offender behaviour and sexual offender programmes and hold governors to account for ensuring these targets are met. 9. The Ministry of Justice should work closely with the National Probation Service and Community Rehabilitation Companies to better understand the relationships between employment, education, settled accommodation and reoffending. As part of this process, they should commission a quantitatively robust evaluation in order to establish a clear evidence base for future rehabilitation strategies. 10. The Ministry of Justice should revisit the contracts it holds with providers under the Transforming Rehabilitation programme in order to enable Community Rehabilitation Companies to better tailor their resettlement service in the light of need and effectiveness. 6

Introduction Which is the best performing prison in the country? Which is the prison that is achieving the best reoffending results?. The answer is we don t know. Seriously we have no idea. This just isn t good enough. 11 Rt Hon David Cameron MP, February 2016 Improving productivity is a key priority for the Government. In the Autumn Statement the Chancellor argued that without sustained productivity improvements the United Kingdom will fail to maintain living standards and achieve economic growth in the longer-term. 12 The Government s 2015 Productivity Plan also stated that greater productivity is essential for the wellbeing of citizens and laid out reforms to the tax, education and transport systems, among others, which aim to support higher levels of productivity. 13 This focus must be extended to public services. The public sector accounts for 20 per cent of GDP, yet only two pages out of 82 were devoted to this topic in the Productivity Plan. 14 Official estimates also suggest public sector productivity has risen only 0.1 per cent per year between 1997 and 2013. 15 The prize for identifying mechanisms for improvement in the public sector alone is therefore sizeable. As Reform argued in Towards a more productive state, however, the current measurement of public sector productivity is poor. 16 By considering only how public spending is translated into outputs, and not longer-term outcomes, current metrics fail to highlight whether organisations are delivering the services that citizens need. A focus on sectors rather than institutions has also hampered the ability for examples of best practice to be identified. Nowhere is this better exemplified than within the prison estate. For years the productivity of the prison service has been measured simply by the amount spent and the number of prisoners in the estate. Publically available data also fails to provide even sector level estimates, but instead combines the prison service with the courts, probation and the fire service. 17 This tells us little about how effectively the public are being protected and prisoners humanely housed. It also fails to capture whether prisons are rehabilitating offenders to desist from committing further crimes. Other mechanisms for performance measurement exist, such as the surveys and inspections carried out by Her Majesty s Inspectorate of Prisons (HMIP) and the annual prison ratings produced by the National Offender Management Service (NOMS) which are informed by prisons reporting on a number of measures. Even these, however, are failing to measure outcomes at an institutional level. The Prime Minister has recognised this challenge. In his recent speech he stated a lack of rigorous performance evaluation is hindering further prison reform and limiting transparency. 18 The Justice Secretary, Rt Hon Michael Gove has also told the Justice Select Committee that the Ministry of Justice (MoJ) is failing to use data effectively to shape policy. 19 Without meaningful performance metrics policymakers and practitioners simply cannot begin to grasp what works and, conversely, what is failing within our current penal system. 20 Key to achieving this will be a move away from what can be easily measured to more innovative ways of capturing success and failure. This will inform the Prime Minister s new prison league tables so that prison leaders can be better held to account. 11 David Cameron, Prison Reform: Prime Minister s Speech. 12 HM Treasury, Spending Review and Autumn Statement 2015. 13 HM Treasury, Fixing the Foundations: Creating a More Prosperous Nation, 2015. 14 Ibid. 15 Office for National Statistics, Public Service Productivity Estimates, 2016. 16 Crowhurst, Finch, and Harwich, Towards a More Productive State. 17 Office for National Statistics, Sources & Methods for Public Service Productivity Estimates: Total Public Services, 2016. 18 David Cameron, Prison Reform: Prime Minister s Speech. 19 Michael Gove, Oral Evidence to the House of Commons Justice Select Committee, 16 March 2016. 20 Chris Fox and Kevin Albertson, Payment by Results and Social Impact Bonds in the Criminal Justice Sector: New Challenges for the Concept of Evidence-Based Policy, Criminology and Criminal Justice 11, no. 5 (August 2011). 7

Unlocking prison performance / Introduction Encouragingly, the Prime Minister s proposed reforms will mean greater use of outcome data such as reoffending and levels of employment post-release. Reducing reoffending remains a key challenge for the prison service and for government. Current evidence suggests nearly half of all prisoners will reoffend within 12 months of release, rising to 60 per cent for those serving less than a year. 21 This revolving door of crime comes at a significant cost to victims and is estimated to cost the public 13 billion a year. 22 Research has also shown that around three quarters of offenders are jobless on release. 23 The inclusion of outcome measurements is therefore essential. This paper proposes a new model for measuring prison performance one which is focussed not only on ensuring decent living conditions for inmates, but also assesses the ability of our prisons to improve inmates life-chances after prison, which is of wider value to society. By ranking prisons based upon a combination of these metrics, policymakers will be better able to assess where value for money is being achieved and subsequently how to spread best practice across the prison estate. The success of such a model is, however, determined by the integrity and quality of the available data, which is a substantial challenge within this area. 8 21 Ministry of Justice, Proven Reoffending Statistics: April 2013 to March 2014, 2016. 22 National Audit Office, Managing Offenders on Short Custodial Sentences, 2010. 23 Ministry of Justice, National Offender Management Service Annual Report 2014/15: Management Information Addendum, 2015.

1 A review of the estate 1.1 The population 11 1.1.1 Offence mix 12 1.1.2 Demographic change 12 1.2 Spending 13 1.3 The estate 14 1.3.1 New for old 14 1.3.2 An estate in crisis 15 1.4 Cost vs quality: assessing performance in the long term 16 1.4.1 Improving rehabilitation 16 9

1 Unlocking prison performance / A review of the estate There has been an ongoing debate on whether prison works. For more than three decades political leaders have held contrasting views about whether prison can reduce reoffending or simply makes bad people worse. This shifting political narrative has had significant implications for prison policy, sentencing frameworks and the headline prison population. Figure 1: Timeline of prison policy affecting headline population, 1990-2015 1992: HMP Wolds becomes the first private prison in England and Wales. 2000: The responsibility for prison healthcare moves from the MoJ to the Department of Health. The Crime and Disorder Act 1998 introduces enhanced sentencing powers for racially or religiously aggravated offences. 1990 1995 2000 The Crime (Sentences) Act 1997 means offenders will now receive an automatic life sentence for a second serious sex or violent offence. Policy announcement New statutory provisions * this timeline is not meant as an exhaustive list While there is debate around who should be imprisoned and for how long, deteriorating prison conditions mean the Justice Secretary is faced with the more pressing question of how to make prison work for those already incarcerated. Recent policy developments have been positive. Since taking up post, Gove has launched reviews into prison education and the juvenile estate, both with a view to improving rehabilitation. 24 Estate modernisation has also been prioritised, again with the idea of improving prisoner outcomes. 25 10 24 Michael Gove, Education in Prison, Press release, (8 September 2015). 25 HM Treasury and Ministry of Justice, Prison Building Revolution Announced by Chancellor and Justice Secretary, 9 November, 2015.

1 Unlocking prison performance / A review of the estate 2010: Justice Secretary, Ken Clarke, publishes his Green Paper: Breaking the Cycle, Effective Punishment, Rehabilitation and Sentencing of Offenders'. 2015: New Justice Secretary, Michael Gove, announces review of education provision in prisons. The Criminal Justice Act 2003 introduces a more comprehensive structure for community sentences. 2012: Justice Secretary, Chris Grayling, announces the rehabilitation revolution but also states prison works, departing from the tone of his predecessor. 2005 2010 2015 The Criminal Justice Act 2003 introduces Indeterminate Sentences for Public Protection (IPP). The Legal Aid, Sentencing and Punishment of Offenders Act 2012 abolishes IPP sentences. 2008: Justice Secretary, Jack Straw, calls for a more punitive approach to offenders and a greater focus on victims. 2013: The MoJ announces seven prison closures and two partial prison closures as part of a new for old estate strategy. Whilst these initial steps should be welcomed, delivering better outcomes from the prison estate means understanding what works. An essential starting point must, therefore, be a better understanding of prison performance. 1.1 The population Between 1993 and 2008 the male prison population grew at a rate of around four per cent annually. 26 This increase can, at least in part, be attributed to a rise in the number of 26 Ministry of Justice, Prison Population Figures: 2014, 2015. 11

1 Unlocking prison performance / A review of the estate prisoners recalled to prison whilst out on licence. In 1999-2000 around 1,300 were recalled, in 2007-08 this was more than ten times higher at just over 13,000. 27 The introduction of legislation which focussed on harsher punishment, including the Crime and Disorder Act 1998 and The Crime (Sentences) Act 1997, also contributed to this rise. 28 In particular, IPP sentences, brought in under the Criminal Justice Act 2003, allowed courts to not only impose a minimum tariff to be served, but also to detain offenders indefinitely until it was possible to prove they posed no risk to the public. Whilst IPP sentences were designed to be reserved for the most serious offenders, they were used much more widely than anticipated by ministers. 29 Human rights concerns combined with the fact that the prison service was ill-equipped to provide the necessary volume of rehabilitation programmes, led to IPP sentences being abolished in 2012. 30 However, those sentenced under the previous regime remain subject to its provisions. As of March 2015 there were still 4,600 prisoners serving IPP sentences many of whom have served their minimum tariff. 31 While those that remain may do so due to the continuing risk they pose, the absence of sufficient access to rehabilitation courses is a key area for improvement for prisons. After the steady increase in inmate numbers however, the Coalition years saw a stabilising of the headline prison population. Between 2010 and 2014 the adult male prison population settled at approximately 81,000. 32 1.1.1 Offence mix Across this period the offence mix of the sentenced population has also broadly remained stable. Offenders convicted of violence against the person have consistently been the largest cohort, particularly since 2009. 33 The number of sexual offenders, however, increased over the last Parliament they now account for 16 per cent of sentenced prisoners compared to 13 per cent in 2010. 34 Prolific offenders with more than 15 previous convictions or cautions also continue to dominate the population, accounting for around one third of prisoners. 35 In the year to June 2014, the proportion of first time offenders fell from 13 to 11 per cent, whilst all other categories remained stable. 36 1.1.2 Demographic change Despite the headline population stabilising over the last Parliament, the demographics of the prison population are changing. The proportion of prisoners over the age of 50 has increased from less than 10 per cent in 2010 to nearly 13 per cent in 2014. 37 At the same time the percentage of prisoners aged under 25 has decreased from approximately 29 per cent to 22 per cent. 38 The Justice Select Committee expects this trend to accelerate in the longer term partially due to increasing prosecutions for historical sex offences although currently the over 50 s still only equate to around 11,000 prisoners. 39 12 27 Ministry of Justice, Offender Management Statistics Quarterly: January March 2013, 2013. 28 Between 2004 and 2014 the average sentence length for indictable offences rose from 16.1 to 18.3 months. See Ministry of Justice, Prison Population Figures: 2014. 29 Ministry of Justice, IPP Factsheet, 2011. See also Andrew Selous, Prison Sentences, 15 October 2015, Written Answer 12127 for the most recent statistics on those still held under the IPP regime. 30 See James, Well and Lee vs United Kingdom in which the European Court of Human Rights found that IPP sentences contravened the right to liberty due to a lack of resources, meaning prisoners were unable to prove they were risk free and able to be released. 31 Andrew Selous, Prison Sentences, 15 October 2015, Written Answer 12127. 32 Over the winter months during this period the adult male population rose to 83,000 falling to around 80,000 during the spring and summer. Ministry of Justice, Prison Population Figures: 2014. 33 Ibid. 34 Ibid. 35 Ibid. 36 Ibid. 37 Ibid. 38 Ibid. 39 House of Commons Justice Committee. Older Prisoners, Fifth Report of Session 2013-14. HC 89. (London: Stationery Office, 2013).

1 Unlocking prison performance / A review of the estate Figure 2: Demographic change in prisons in England and Wales 25,000 +9.6% 20,000 Prison population 15,000 10,000-33.5% -12.6% +2.9% +4.6% +35.9% 5,000 +30.6% -55% 0 15-17 18-20 21-24 25-29 30-39 40-49 Age group 2010 2014 % shows the percentage change between 2010 and 2014 50-59 60 and over Source: Ministry of Justice, Prison population, 2014 1.2 Spending Public expenditure on prisons grew rapidly through the late 1990s and mid-2000s, partly due to increases in the headline population. 40 This trend was reversed in 2010 when a tough financial settlement led to a rapid reduction in costs. 41 More recently, the 2015 Spending Review settlement committed 1.3 billion of capital investment to transform the current prison estate. 42 It also requires the MoJ to save a further 930 million by 2019-2020. 43 The proportion of these cuts that will fall on prisons is not yet known, however, the Government has outlined a minimum target of 80 million of savings per year from prison running costs, out of a total budget of approximately 2 billion. 44 Over the last Parliament, estate restructuring measures particularly the closure of old inefficient prisons, combined with financial benchmarking succeeded in bringing down the unit cost of prisons across England and Wales. The total average annual cost of housing a prisoner was 37,163 in 2010-11 falling to 33,785 by 2013-14, a reduction of 9 per cent. 45 There remains, however, significant variation in spending across the estate. For example, amongst Category C prisons in 2014-15, the highest cost per prison place was HMP Kennet at 73,828 per year compared with HMP Oakwood at only 18,549 around a quarter of the cost. 46 40 HM Treasury, Public Expenditure Statistical Analyses 1999-2000, 1999; HM Treasury, Public Expenditure Statistical Analyses 2005, 2005; HM Treasury, Public Expenditure Statistical Analyses 2011, 2011. 41 HM Treasury, Spending Review 2010, 2010. The Ministry of Justice faced cuts of 23 per cent in real terms over the period 2010-2015. 42 Ministry of Justice and HM Treasury, Ministry of Justice s Settlement at the Spending Review 2015, 2015. 43 Ibid. 44 Ibid. This figure refers to the direct resource expenditure on prisons found in table 1 in Ministry of Justice, Costs per Place and Cost per Prisoner by Individual Prison Establishment 2014-15 Tables. 45 Ministry of Justice, Costs per Place and Cost per Prisoner by Individual Prison Establishment 2011-12, 2012. ; Ministry of Justice, Costs per Place and Costs per Prisoner by Individual Prison Establishment 2014-15 Tables. 46 Ministry of Justice, Costs per Place and Cost per Prisoner by Individual Prison Establishment 2014-15 Tables. 13

1 Unlocking prison performance / A review of the estate 1.3 The estate The prison estate in England and Wales is managed by NOMS, which is an executive agency of the MoJ. There are some 121 establishments. Currently 14 of those are privately run. The majority of prisons are categorised according to the level of security they provide. 47 Since the late 1960s adult male offenders have been classified as either Category A, B, C or D depending on the perceived likelihood of them attempting to escape and the risk of harm to the public of such an attempt. 48 The least serious offenders, and those reaching the end of longer sentences whose risk-level is deemed to have reduced, are housed in open prisons with higher levels of autonomy and freedom of movement. High-security prisons hold the more serious Category A and Category B prisoners who are serving long sentences. In England and Wales, a long sentence is defined as one for which the threshold for conditional early release is over four years. 49 Throughout the duration of their sentence, offenders can move between categories depending on their perceived security risk to the public and their engagement with prison programmes. Whilst being detained before trial, or immediately following sentencing, the majority of prisoners are sent to local prisons, after which they may be transferred to longer-term accommodation. 1.3.1 New for old The current estate varies widely in age and design. A significant number of prisons were built over a century ago, whilst only a minority were built in the last decade such as HMP Oakwood, which opened in 2013. 50 As a result, a number of prisons have limited space for rehabilitative activities, are difficult to modernise and are costly to run. 51 Successive governments have attempted to solve this problem. In 2006, the Labour Government announced the creation of the Core Capacity and New Prisons Programmes. 52 Central to these reforms was a commitment to increase the net capacity of the prison estate to 96,000 by 2014, following a period of population influx. While rising prison numbers provided a burning platform, the then Government also saw modernisation of the estate as paramount. The New Prison Programme in particular specifically aimed to replace over five thousand worn out, inefficient places in the current estate. 53 Lord Carter s review of prisons also proposed Titan prisons, large institutions able to hold up to 2,500 prisoners. 54 The review argued that increasing capacity in this way would maximise the purposeful activity available to prisoners whilst also allowing a large number of smaller, older and less efficient prisons to be closed. Concerns were expressed, however, that Titan prisons would prove difficult to run and result in less humane regimes. 55 Her Majesty s Chief Inspector of Prisons (HMCIP) at the time also believed that smaller institutions could provide more suitable conditions for rehabilitation. 56 As a result of significant opposition to the MoJ s consultation paper on Titan prisons the plans were abandoned. 57 Subsequent analysis of prison performance has found that newer prisons, regardless of their size, outperformed older institutions. 58 14 47 Ministry of Justice, Categorisation and Recategorisation of Adult Male Prisoners, 2011. 48 Ibid. 49 Andrew Coyle, The Management of Prisoners Serving Long Sentences, (International Centre for Prison Studies, 2001). 50 Prisons that were built over a century ago include HMP Dartmoor, HMP Pentonville and HMP Wandsworth. 51 Lord Carter, Securing the Future: Proposals for the Efficient and Sustainable Use of Custody in England and Wales, 2007. 52 Gabrielle Garton Grimwood, Building Prisons: The Bigger, the Better? (London: Stationery Office, 2014). 53 Ibid: 17. 54 Lord Carter, Securing the Future, 2007. 55 Centre for Social Justice, Locked Up Potential: A Strategy for Reforming Prisons and Rehabilitating Prisoners, 2009. 56 HM Inspectorate of Prisons for England and Wales. HM Chief Inspector of Prisons for England and Wales Annual Report 2008-09, 2009. 57 Gabrielle Garton Grimwood, Building Prisons: The Bigger, the Better?. 58 Kevin Lockyer, Future Prisons: A Radical Plan to Reform the Prison Estate (Policy Exchange, 2013).

1 Unlocking prison performance / A review of the estate The Coalition Government also sought to modernise the estate through a programme of restructuring. As with previous reforms, this involved closing older, more expensive prisons and investing in new, cheaper capacity a strategy of new for old. Since 2010, 14 prisons have been closed and two new prisons have opened. 59 The Government has also signed a contract to build a 2,100 capacity prison in Wrexham a partial return to Lord Carter s Titan prisons. 60 The National Audit Office (NAO) described the programme as the most coherent and comprehensive for many years. 61 In late 2015 Justice Secretary, Rt Hon Michael Gove, pledged to continue this programme, announcing the closure of a further nine old prisons, with five new prisons set to be built before the end of this Parliament. The new for old strategy is therefore forming a key plank of the Conservative Government s prison reform agenda. 1.3.2 An estate in crisis Despite this programme of modernisation, and the reforms to rehabilitation services, it has consistently been argued that prisons are failing in one of their most basic requirements of delivering a humane environment for offenders. In his final annual report previous HMCIP, Nick Hardwick, found that outcomes fell sharply across all areas and, overall, the outcomes we reported on in 2014 15 were the worst for 10 years. 62 Similarly, the Justice Select Committee have argued that aside from HMIP, evidence from the Independent Monitoring Boards (IMBs), MoJ and Prison and Probation Ombudsman (PPO) all indicate a deterioration in standards of safety and performance across the prison estate. 63 Violence Over the last five years there have been worrying increases in assaults and violent incidents, both amongst inmates and against staff. The total number of assaults has risen from 14,508 in 2010, to 18,874 in 2015. 64 Assaults against staff have risen from 2,937 to 4,568 over the same time period. 65 2015 also saw the largest number of homicides (eight) recorded in a single calendar year. 66 The number of incidents of self-harm and self-inflicted deaths are also a cause for concern. Whilst the number of self-inflicted deaths remained stable in 2014 and 2015, the longer-term trend shows a steady rise, increasing from 58 in 2010 to 89 in the year ending March 2014 and in the context of a relatively constant population size. 67 One contributing factor to the increase in prison violence is the increasing prevalence of new psychoactive substances dubbed lethal highs following a number of inmate deaths resulting from their use. 68 The last two annual reports from the HMIP have also raised concerns about this issue, with their 2014-15 report stating: [p]risoner violence towards staff and other prisoners had risen, often fuelled by the increased use of new psychoactive substances. 69 59 HMPs Thameside and Oakwood. 60 Ministry of Justice, Modernisation of the Prison Estate, Press Release, (4 September, 2013). 61 National Audit Office, Managing the Prison Estate, 2013: 9. 62 HM Inspectorate of Prisons for England and Wales, HM Chief Inspector of Prisons for England and Wales Annual Report 2014-15, 2015: 7. 63 House of Commons Justice Committee, Prisons: Planning and Policies, Ninth Report of Session 2014 15 (House of Commons, 2015): 3. 64 Ministry of Justice, Safety in Custody Statistics England and Wales Deaths in Prison Custody to September 2015 Assaults and Self-Harm to June 2015, 2015. These figures include incidents that occurred at Immigration Removal Centres. 65 Ministry of Justice, Safety in Custody Quarterly Update to September 2015. 66 Ibid. 67 Ibid. 68 Andrew Selous, Safety in Prisons for singles, HC Deb 17 June 2015, c 81WH. 69 HM Inspectorate of Prisons for England and Wales, HM Chief Inspector of Prisons Annual Report 2014-15, 2015: 32. 15

1 Unlocking prison performance / A review of the estate Overcrowding 70 Despite the headline population stabilising, overcrowding levels have continued to rise. The most recent figures for 2014 show an increase in the number of prisons which were classified as overcrowded from 77 of 119 prisons in March, to 83 of 117 prisons in December. 71 In June 2015, the MoJ also admitted that previous statistics going back to 2009 had underestimated actual levels of overcrowding. 72 Inaccurate recording practices meant that in some cases two prisoners being held in a cell designed for one had only been counted as a single instance of overcrowding. Revised figures taking these inaccuracies into account therefore suggest a higher level of overcrowding than previously thought. 73 For example, in 2013-14 the original figures put overcrowding at 22.9 per cent and the number of offenders doubled up in a cell designed for one at 21.9 per cent. 74 Revised data showed this to be in fact 24.1 and 24.5 per cent respectively. 75 1.4 Cost vs quality: assessing performance in the long term With budgets falling over the last Parliament the prison service has seen an increasing pressure to cut costs and quickly. The above information shows that as a result, there has been a growing tension between short-term efficiency and longer-term effectiveness. The Justice Select Committee has argued that it is improbable that there is no link between estate reconfiguration, benchmarking, and changes in operational policy and the shift in safety across the prison estate. 76 In particular, it suggests that reductions in staffing numbers, due both to spending cuts and increased staff vacancies, have resulted in a more restrictive regime in a number of institutions, undermining relationships between prisoners and staff. 77 A previous analysis of short-sentenced prisoners also argued that overcrowding can lead to reduced provision of activity for prisoners which can be harmful in the long term. 78 In addition, the NAO have expressed concerns that the Coalition s estate strategy led to a number of high performing prisons being closed. Despite being projected to save 211 million between 2010 and 2015 just under half of the prisons closed, or identified for closure, were considered high-performing scoring 13 or more out of a possible 16 in their most recent inspections. 79 While, as argued in this paper, current mechanisms for measuring success fall short, particularly by not evaluating outcomes, it is still concerning that overall prison performance was not considered as part of the decision-making process. High-performing prisons that are expensive to run are not necessarily desirable, but, using some measure of prison performance is vital to ensure taxpayers money is spent to the best effect both in the short and long term. 1.4.1 Improving rehabilitation Aside from maintaining service levels, value for money is most importantly about delivering the best possible outcome for the lowest possible price. A crucial, and significant part of this is assessing a prison s ability to reform the prisoners it houses. Over the last decade, however, the prison estate has made little headway in decreasing the number of individuals who go on to commit further offences once released from prison. 16 70 There are two key definitions when considering whether a prison is overcrowded: (i) Certified Normal Accommodation (CNA) also known as uncrowded capacity, is the Prison Service s own measure of accommodation. CNA represents the good, decent standard of accommodation that the Service aspires to provide all prisoners. (ii) Operational capacity is the total number of prisoners that an establishment can hold without serious risk to good order, security and the proper running of the planned regime. Ministry of Justice, Certified Prisoner Accommodation, 2012. 71 House of Commons Justice Committee, Prisons: Planning and Policies, Ninth Report of Session 2014-15, 2015. 72 Andrew Selous, Written Statement HCWS29, 11 June 2015. 73 Ibid. 74 Ibid. 75 Ibid. 76 House of Commons Justice Committee, Prisons: Planning and Policies, Ninth Report of Session 2014-15, 2015: 4-5. 77 Ibid. 78 National Audit Office, Managing Offenders on Short Custodial Sentences, 2010. 79 A total of 8 out of 18 prisons achieved these scores. National Audit Office, Managing the Prison Estate.

1 Unlocking prison performance / A review of the estate The latest available data from the financial year 2013-14 records the adult reoffending rate for those sentenced to custody at 46 per cent. 80 Such high levels come at a significant cost to both offenders and society. At a national level the NAO have estimated that reoffending by all recent ex-prisoners costs the economy between 9.5 billion and 13 billion per year. 81 While these estimates are taken from the year 2008-09 the fact that the reoffending rate has remained almost constant over the last decade means these cost are unlikely to have fallen and if anything would have increased in line with inflation. The most worrying offender category remains those serving sentences of less than 12 months. Since 2002, the reoffending rate for this cohort has consistently been much higher than that of those serving longer sentences. Despite nearly a 2 per cent reduction in the year to 2013, their reoffending rate was 58 per cent, compared to 34 per cent for those serving more than 12 months. 82 From 2013, the Coalition Government s Transforming Rehabilitation (TR) programme extended resettlement support and supervision to prisoners serving sentences of less than 12 months. It remains too early to assess the impact of this change. 83 80 Ministry of Justice, Proven Reoffending Statistics: April 2013 to March 2014. 81 National Audit Office, Managing Offenders on Short Custodial Sentences: 4. 82 Ministry of Justice, Transforming Rehabilitation: A Strategy for Reform, 2013. 83 Ibid. 17

2 Measuring performance in prisons 2.1 Why outcomes count 19 2.1.1 Isolating the prison effect 19 2.1.2 Exploring variation 20 2.2 Current methods 21 2.2.1 Outcome or output? 23 2.3 The way forward 24 18

2 Unlocking prison performance / Measuring performance in prisons The deterioration in safety across the estate, combined with a continuing failure to reform offenders highlights the importance of balancing short-term financial gains with longerterm objectives. Critical to achieving a better understanding of prison performance is ensuring evaluations consider outcomes and not simply the volume of prisoners housed or the quality of the prison environment. This chapter will outline the importance of an outcomes-focussed approach and consider a number of measurement challenges within the justice setting. It will also lay out current mechanisms for understanding prison performance, before proposing a new model which aims to ensure value for money is maintained in the short, medium and longer term. 2.1 Why outcomes count As the Prime Minister has recently argued, demonstrating the value of our public services requires us to measure what really counts in this case whether our prisons are equipping offenders with the skills they need to reintegrate into society and desist from crime. 84 This alone makes it an essential part of Reform s model for measuring prison performance. However, there are also wider societal benefits from tackling reoffending. Firstly, the prison population is disproportionately disadvantaged when compared to the general population. In a survey of 1500 offenders, 24 per cent had been in care during childhood and 29 per cent had experienced abuse. 85 This compares with 2 and 4 per cent, respectively, in the general population. 86 Multiple studies, including the 2009 Bradley Review, also suggest significant numbers within the prison population suffer with one or more recognised mental illness. 87 Preventing further incarceration therefore helps to improve the wellbeing of the least fortunate. Studies also show that victims, just like offenders, are predominantly found in disadvantaged communities. 88 The Office for National Statistics (ONS) has found victimisation to be consistently associated with lower levels of personal wellbeing meaning reducing reoffending has benefits beyond the prisoner population. 89 Increasing the proportion of prisoners that are living in settled accommodation and are in education or employment after release is important not just for prisoners wellbeing but for the long-term sustainability of government finances. These post-release outcomes help to create better skilled, more employable citizens, which in the right economic environment might lead to reduced reliance on the welfare state and thus more sustainable public spending. In short, an outcomes-focussed prison system which also ensures prisons are spending wisely, benefits prisoners, victims and wider society. 2.1.1 Isolating the prison effect Disentangling the impact of external factors to isolate the so-called prison effect on outcomes, such as reoffending and employment upon release, is complex. The prison estate forms only part of a complicated web of government agencies. It is challenging, for example, to separate out the role of the Prison Service from probation or Integrated Offender Management (IOM) teams, which are also charged with reducing offending. The relationship between the prison and probation services is particularly pertinent in light of 84 David Cameron, Prison Reform: Prime Minister s Speech. 85 Kathryn Hopkins, Wave 1 of Surveying Prisoner Crime Reduction (SPCR), a Longitudinal Cohort Study of Prisoners Conducted from 2005 to 2010 in England and Wales, (Ministry of Justice, 2013). 86 Ibid. 87 Lord Bradley, The Bradley Report (Department of Health, 2009). 88 Those living in the 20 per cent most deprived areas are more likely to be victims of crime. Office for National Statistics, Crime Statistics, Focus on Public Perceptions of Crime and the Police, and the Personal Well-Being of Victims, 2013 to 2014, 2015. 89 Ibid. 19

2 Unlocking prison performance / Measuring performance in prisons the TR programme. 90 This saw the creation of through the gate services which aimed to join-up the support received by offenders under new Community Rehabilitation Companies (CRCs). 91 In addition, offenders, particularly those serving sentences over 12 months, are likely to have been housed in more than one establishment. It can therefore be difficult to understand whether improvements or declines in prisoner outcomes can be attributed to the prison from which they were released or whether the other establishments in which they were housed had the most impact on their propensity to reoffend. Work by the Home Office went some way in attempting to isolate the impact of a particular prison by comparing the seven-year reconviction rate of prisoners who attended HMP Grendon with a control group (who met the same selection criteria, and had similar risk levels) who were sent to an alternative establishment. 92 By attempting to emulate the design of a randomised control trial the researchers were able to control for the inherent characteristics of prisoners which was particularly important at HMP Grendon where a large number of prisoners suffered from personality disorders and were therefore considered high-risk. 93 The results suggest that time at HMP Grendon was strongly related to lower reoffending rates, compared to time at a different prison. 94 For prisoners spending more than 18 months at the establishment this impact was more pronounced. It is of course important to note that HMP Grendon is a specialist prison run as a therapeutic community and therefore it is perhaps expected that outcomes would be different from non-specialist prisons. The study is also based on data over two decades old. In recent times few studies have been able to address these issues. Similar to other public services, background characteristics also have a large impact on the chances of an individual reoffending. Evidence suggests that a prisoner s gender, age and offending history have a much greater effect on an individual s propensity to reoffend than criminal justice interventions such as vocational training. 95 For example, results from the Surveying Prisoner Crime Reduction (SPCR) longitudinal study show that criminal history has the strongest association with further offending independent of other factors including crime type, employment status prior to prison and substance misuse. 96 While data on age and offence history may be available at a national or regional level, no information (aside from gender due to prison categorisation) is made publically available at a prison or prisoner level. Some of this data is held on the Police National Computer (PNC) and can be utilised for assessing risk under integrated offender management schemes. For the vast majority of researchers, however, an inability to access this resource provides an insurmountable barrier to controlling for these prisoner characteristics thus making it difficult to isolate the prison effect from a purely quantitative perspective. 97 2.1.2 Exploring variation Whether or not prisons have a comparatively low impact on reoffending rates relative to these other factors an offender s prison experience remains an important part of the rehabilitation puzzle. Statistics on international performance shows wide variations in 20 90 Measuring performance using an Administration of Justice approach, which takes into account these inter-relationships would be one way to acknowledge this overlap. See Office for National Statistics, Public Service Productivity, Measuring the Output of the Probation Service, 2010. 91 While this may be an important challenge for researchers and policymakers moving forwards, the data used in this paper is taken from the financial year 2012-13 and therefore will not be affected by the changes introduced by Transforming Rehabilitation. 92 Ricky Taylor, Seven-Year Reconviction Study of HMP Grendon Therapeutic Community (Home Office, 2000). 93 Ibid. 94 Ibid. 95 Ian Brunton-Smith and Kathryn Hopkins, The Factors Associated with Proven Re-Offending Following Release from Prison: Findings from Waves 1 to 3 of Surveying Prisoner Crime Reduction, (Ministry of Justice, 2013). 96 Ibid. International evidence also supports this claim. Work by the State of Colorado found that the characteristics of prisoners, such as age had a significant effect on recidivism rates. For example, offenders under 20 had a reoffending rate of 53 per cent compared to 41 per cent for the 50-59 category, see Ryan King and Brian Elderbroom, Improving Recidivism as a Performance Measure (Justice Policy Center, 2014). 97 For some academic access to a sample of PNC data may be possible. All data in these instances would be anonymised.